diff --git a/parsel/selector.py b/parsel/selector.py
index 2027599..a5db974 100644
--- a/parsel/selector.py
+++ b/parsel/selector.py
@@ -423,6 +423,7 @@ class Selector:
"_huge_tree",
"root",
"_text",
+ "_text_lazy_html_root",
"body",
"__weakref__",
]
@@ -507,6 +508,7 @@ def __init__(
self._expr = _expr
self._huge_tree = huge_tree
self._text = text
+ self._text_lazy_html_root: Optional[etree._Element] = None
def __getstate__(self) -> Any:
raise TypeError("can't pickle Selector objects")
@@ -606,7 +608,12 @@ def xpath(
)
else:
try:
- xpathev = self._get_root(self._text or "", type="html").xpath
+ if self._text_lazy_html_root is None:
+ self._text_lazy_html_root = self._get_root(
+ self.root or "", type="html"
+ )
+ if self._text_lazy_html_root is not None:
+ xpathev = self._text_lazy_html_root.xpath
except AttributeError:
return typing.cast(
SelectorList[_SelectorType], self.selectorlist_cls([])
@@ -722,25 +729,30 @@ def get(self) -> Any:
For HTML and XML, the result is always a string, and percent-encoded
content is unquoted.
"""
- if self.type in ("text", "json"):
+ if self.type in ("json", "text"):
+ if self.type == "text" and self._text_lazy_html_root is not None:
+ return etree.tostring(
+ self._text_lazy_html_root, encoding="unicode", with_tail=False
+ )
return self.root
- try:
- return typing.cast(
- str,
- etree.tostring(
- self.root,
- method=_ctgroup[self.type]["_tostring_method"],
- encoding="unicode",
- with_tail=False,
- ),
- )
- except (AttributeError, TypeError):
- if self.root is True:
- return "1"
- elif self.root is False:
- return "0"
- else:
- return str(self.root)
+ else:
+ try:
+ return typing.cast(
+ str,
+ etree.tostring(
+ self.root,
+ method=_ctgroup[self.type]["_tostring_method"],
+ encoding="unicode",
+ with_tail=False,
+ ),
+ )
+ except (AttributeError, TypeError):
+ if self.root is True:
+ return "1"
+ elif self.root is False:
+ return "0"
+ else:
+ return str(self.root)
extract = get
diff --git a/tests/test_selector.py b/tests/test_selector.py
index 96713f9..902f82b 100644
--- a/tests/test_selector.py
+++ b/tests/test_selector.py
@@ -1007,6 +1007,35 @@ def test_remove_selector(self) -> None:
self.assertIsSelectorList(sel.css("li"))
self.assertEqual(sel.css("li::text").getall(), ["2", "3"])
+ def test_remove_selector_from_html_in_text(self) -> None:
+ html = (
+ "hello world
"
+ )
+ expect_result = "hello world
"
+ sel = self.sscls(text=html, type="text")
+ self.assertEqual(sel.type, "text")
+ li_sel_list = sel.css("style")
+ li_sel_list.drop()
+ self.assertEqual(sel.get(), expect_result)
+ # The type of the parent selector should not change
+ self.assertEqual(sel.type, "text")
+
+ def test_remove_selector_from_html_in_json(self) -> None:
+ json_str = """{
+ "title": "hello world",
+ "body": "hello world
"
+ }
+ """
+ expect_result = "hello world
"
+ sel = self.sscls(text=json_str)
+ html_sel = sel.jmespath("body")[0]
+ self.assertEqual(html_sel.type, "text")
+ li_sel_list = html_sel.css("style")
+ li_sel_list.drop()
+ self.assertEqual(html_sel.get(), expect_result)
+ # The type of the parent selector should not change
+ self.assertEqual(html_sel.type, "text")
+
def test_remove_pseudo_element_selector_list(self) -> None:
sel = self.sscls(
text=""